earnings anticipated from it. Lifetime earnings associated with additional education are projected from age-earnings profiles by years of school at a given point in time. Some of the more basic theoretical problems with this approach are the following:* 1. Returns to society are normally computed from returns to individuals who receive them, but there is in fact little evidence on the relationship between the two. 2. It is often very difficult anyway to disentangle the private returns from education in a society in which educational opportunity is closely correlated with wealth and status and in which " 'good breeding' earns more . . . than a good education" (69, p. 45, fn. 1), and this covers many Latin American and Asian countries, though not typically those of tropical Africa. 3. In this as well as in other respects, labor markets in the developing countries are normally very imperfect, and earnings do not, therefore, adequately reflect the relative scarcities of people with different levels of education and skill. This problem can to some extent be overcome by using shadow wages—employing accounting wages for planning purposes which do reflect the relative scarcities of different kinds of labor. 4. In projecting private rates of return from education it is often (probably normally) invalid to assume a close relationship between recent rates of return and future rates of return. The link would appear to be particularly tenuous in a less developed country in which differences in earnings between the better and lesser educated are very wide at the outset because of extreme scarcity of skilled personnel. In such a country differences may either narrow considerably as education becomes more widely disseminated or possibly widen if rapid economic growth boosts the demand for educated personnel, and expansion of education does not keep pace. 5. The developing countries are concerned not with marginal changes, but with massive structural changes, not only in the economy in general but in the education system itself. In such a situation it is extremely difficult to forecast rates of return on investments in education even a few years ahead, since they will be partly a function of the overall pattern of investment that is adopted and the flowback effects that occur. As shown earlier, there are a number of such flowback effects between education and fertility trends. Returns from Family Planning Programs The kinds of problems outlines in 4 and 5 apply as well to the analysis of family planning programs, but the application of rate-of-return analysis to these programs faces some additional special problems. *For a more detailed analysis of some of these problems, see (69, Ch. Ill; 70; 71, pp. 19-30).t in recent years (57). The very low percentages in Singapore, Taiwan, and Turkey may partly reflect a high standard of public health care, and partly (especially in Singapore and Turkey) the fact that family planning expenditures are still increasing as the programs gain momentum. The high percentage in Korea reflects a remarkably small public health budget rather than a large family planning budget.